A fundamental issue in speech research is the interaction between production and perception. The nature of this interaction has profound implications for understanding and modeling speech development, production deficits, and rehabilitation strategies. Our goal is to characterize how perception of others' speech, particularly at the phoneme boundary, influences the auditory-motor feedback processes that guide self- produced speech. On the one hand, a speaker may perceive his own speech in the categorical manner in which listeners perceive it, allowing for rapid, robust auditory processing. On the other hand, a speaker may monitor his output at a sub-categorical level, before high-level auditory cortex imposes phonetic structure on the acoustic signal. We aim to distinguish these hypotheses by examining speech under conditions of auditory change and probing the neural signal for an increased response to that change when it is phonetically relevant. Our project combines psychophysical and magnetoencephalography experiments to investigate the neural dynamics elicited by a sudden modification of speakers' auditory feedback. The proposed experiments were designed to achieve two specific aims. First, we aim to measure the neural responses to real-time phonetic category changes. Drawing on the results of past studies, we hypothesize that a cross-category or phonetic shift causes a greater neural response than a within-category or non-phonetic shift of the same magnitude, but the dynamics of this response are still unknown. The goal is to use MEG to examine the time- varying neural response to unexpected feedback perturbation, contrasting that response under conditions of phonetic and non-phonetic change. Secondly, we aim to assess the effects of speech training on auditory feedback control. Learning a novel vowel target in formant space has the effect of adding new category boundaries between the novel vowel and the well-learned native vowels. The proposed experiments evaluate the degree to which these newly-learned categories affect the responses to perturbation of an existing vowel. The proposed research adds to the existing feedback literature by introducing the distinction between meaningful linguistic changes and mere acoustic variations imposed in feedback. We aim to improve models of speech motor control by determining whether auditory feedback control is influenced by categorical perception, and therefore whether it occurs at a high or low level in auditory cortex. This research is directly applicable to stuttering, a motor control disorder thought to reflect abnormalities in feedback processing. These studies will also ultimately contribute to improved diagnosis and treatment of communication disorders such as Parkinsons' Disease or spasmodic dysphonia, since neuroimaging of feedback control can be used diagnostically to probe the specific abnormalities in brain networks involved in perception and production. Finally, the training studies proposed here could potentially be useful in developing feedback-related training strategies for a variety of speech disorders.